Open lindeloev opened 7 years ago
@lindeloev
Thanks for this suggestion. At the moment we are at the in the middle of a new release 0.8.3 (with logistic regression), but we will really consider this enhancement in a near future release, because we agree that this suggestion can be useful especially for educational purposes. In general improving the plots is on our list (together with #1983 ), because we are aware that there are limitations there.
Kind regards Frans Meerhoff
Thanks! I'm really excited about your upcoming release. I think it will be the one that allows me to finally ban SPSS from teaching.
Isnt this something that is partly solved by the distributions module? @Kucharssim? I think HPD intervals are missing though.
Yes it is partly addressed by the Distribution module, but I do think there could be value to a new plot option that just shows the prior (this is done in the Learn Bayes module, for instance). The question is where this ranks on our list of priorities...but I do like the general idea.
Enhancement: Ticking off "prior and posterior" should plot the prior distribution without data. An additional option could show prior HPD ("Highest Prior Density"?) intervals. Currently, no plot is shown until data is entered.
Purpose: Bayesian inference sets prior(s) before updating beliefs from data. If JASP allowed for this, it would be great in multiple ways, though I'm mostly raising this for educational purposes. An HPD interval is particularly useful for wide priors where probability mass is difficult to extract visually. Solving https://github.com/jasp-stats/jasp-desktop/issues/1983 down the road will also be helpful here.